[openbiblio-dev] [pd-discuss] Bibliographic Metadata Guide

Wed Sep 7 16:23:46 UTC 2011

I'm not at all sure about the section on metadata data models in:
   http://openbiblio.okfnpad.org/metadata-art

It seems to be a mix of models and serializations. You might want to look at:

http://dublincore.org/documents/singapore-framework/

in particular the 3-layer diagram. In that framework, you have domain  
models which I find to be very useful ways of thinking about the  
metadata. For example, libraries have an older domain model,  
International Standard Bibliographic Description, and a newer one,  
Functional Requirements for Bibliographic Records. Those describe the  
entities of the domain.

RDF is an even lower level model. I don't know of anything that rivals  
RDF at that level.

For serializations you have things like turtle (for triples), XML (for  
data that can be marked up in a flat record),  JSON, another  
record-based serialization. Even MARC is a serialization that can  
carry a variety of data types (as ISO 2709).

To me, a model describes the entities of your metadata "realm" and is  
independent of any serialization. So if you want your metadata to  
cover, say, text documents, then you would define your entities:

resources
   independent resources
   contained resources (articles, chapters)
   resource containers (journals, books)
agents
   creators
     persons
     corporate bodies
etc.

An example of this is the Scholarly Works Application Profile:

http://www.ukoln.ac.uk/repositories/digirep/index/Eprints_Application_Profile#Scholarly_Works_Application_Profile

which begins with an E-R model:

http://www.ukoln.ac.uk/repositories/digirep/index/Model

The model itself has the potential to be implemented in various ways  
using different data elements and different data structures. Once the  
model is clear and the data elements have been defined, then you can  
choose one or more serializations. The serialization is really the  
least important part, since most data can be conveyed using more than  
one serialization.

Sorry to go on about this, but I think that using this methodology you  
are less likely to paint yourselves into a corner, something that  
happens rather frequently with metadata. These methods help you create  
metadata that is extensible and that has a solid model as its basis.

kc

Quoting Primavera De Filippi <primavera.defilippi at okfn.org>:

> Dear all,
> given the recent activity on the Bibliographic Metadata Guide, I
> thought it would be nice to clean things up a bit and re-organise the
> whole thing.
> I decided to split it into different sections: the old etherpad is now
> deprecated and has been replaced by the following pads:
>
> - http://openbiblio.okfnpad.org/metadata-toc = Table of content +
> generalities, links, and informations
> - http://openbiblio.okfnpad.org/metadata-art = State of the Art -
> review of the different standards + who uses what
> - http://openbiblio.okfnpad.org/metadata-procons = Analysis of the pro
> & cons of the most popular standards
> - http://openbiblio.okfnpad.org/metadata-conclusions = our conclusions
> - what do we want to propose as the "best" standard(s) for
> bibliographic metadata
>
> It would be great if you can take a look of those pads and make sure
> everything is correct, or perhaps add whatever you thing should be
> mentioned.
> As usual, any comments or feedback are greatly appreciated  :)
> Keep on with the good work !
> Primavera
>
>
>
> On Tue, Sep 6, 2011 at 1:20 PM, Primavera De Filippi
> <primavera.defilippi at okfn.org> wrote:
>> Hi Jim and everoyne,
>> thank you all for you feedback - any comment is greatly appreciated
>> and please do keep contributing !
>> A lot of discussion is currently going on in the Bibliographic
>> Metadata Guide's etherpad: http://okfnpad.org/metadata
>> I think it is important that the community is and remains involved in
>> this discussion because we want to reach a consensus from the
>> community.
>> So if anyone is either interested or concerned by the use of metadata
>> standards in the bibliographic area, take a look at the pad:
>> http://okfnpad.org/metadata
>> the most interesting sections at the moments are: ##Goals, and
>> ##Issues to be addressed
>> any contribution and feedback is welcome   ;)
>> Thank you !
>>
>>
>>
>>
>> On Wed, Aug 24, 2011 at 5:02 PM, Jim Pitman  
>> <pitman at stat.berkeley.edu> wrote:
>>> Primavera De Filippi <primavera.defilippi at okfn.org> wrote:
>>>
>>>> The term "Auto-descriptive Metadata" was indeed unclear, I changed it
>>>> into "Self-descriptive Metadata" - whenever the metadata contains
>>>> sufficient information for the component and its relationship to the
>>>> conference series to be completely self-describing, versus "Non
>>>> Self-descriptive Metadata" - whenever the meaning of the markup
>>>> language is implemented in the logic of the parser, i.e. the metadata
>>>> is not self-descriptive.  Do you think that's more accurate and clear
>>>> ?
>>>
>>>
>>> No!  What does "relationship to the conference series" mean for a book?
>>> What does "completely self-describing" mean?  Why does this distinction
>>> (whatever is intended) make a useful categorization?
>>> Also, in the pad I see something different again:
>>>> Main distinction is between:
>>>> 1. self-descriptive metadata (based on a metadata data model)
>>>> 2. the rest
>>>
>>> The meaning of this distinction is not  clear to me. Take for  
>>> example BibTeX.
>>> This page http://en.wikipedia.org/wiki/BibTeX provides an almost  
>>> machine-readable
>>> description of the BibTeX data schema. Isnt that a metadata data model?
>>>
>>> I see DC is under both 1. and 2.
>>> I am left with no idea what is intended by the distinction or why  
>>> it might be useful.
>>>
>>> --Jim
>>>
>>
>

-- 
Karen Coyle
kcoyle at kcoyle.net http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet